Peer Groups Can Cause You Major Headaches, So Here Is A More Sensible Alternative
Morningstar, Lipper and S&P have all used a headline like this many times. Sometimes it refers to growth managers and sometimes to value managers: The pendulum swings, and I can predict when it happens. After you read this, you’ll be able to predict as well because it’s a peer group problem, not an incompetence issue.
Do you really believe that investment managers can alternate between brilliance and stupidity? Of course not. It’s actually quite simple: When a style is out of favor, the majority of managers in that style will outperform. When the style is in favor, the majority will underperform.
For the last five years the Russell 1000 value index is down -2.6% while the growth index is up 2.5%. The current situation looks like this:
Most importantly, peer groups will mess you up because they are loaded with biases, including one particularly insidious bias that most do not know or understand. It’s called classification bias, and is described in detail at Classification Bias.
In a nutshell, the majority of funds in a peer group, like large value, do not belong in that peer group, so when that style is out of favor, being out of that style is in favor, hence the majority wins (please think about this).
The situation gets much worse with hedge fund peer groups, because for the most part each fund in a hedge fund peer group is not like most of the other funds in that peer group. So funds rank well or poorly because of the misclassification rather than because they have succeeded or failed. Worse, managers get hired and fired for the wrong reasons, and clients suffer.
But all is not lost. There is an alternative to peer groups that is totally unbiased so you and your clients make better decisions, and it’s far less expensive than the universes you’re currently paying for. Wouldn’t you like to pay less for better information?
Performance evaluation is a hypothesis test. We want to accept the hypothesis that performance is good. Crack open those stat books folks, where you’ll find that hypotheses are tested by comparing the actual outcome to all of the possible outcomes. That’s exactly what we’ve done to replace peer groups.
Portfolio Opportunity Distributions (PODs) create all of the portfolios the manager could have held, selecting stocks at random from the manager’s benchmark. You then compare what actually happened to all of the returns that could have happened, and you can be confident that your inferences are not contaminated by the host of biases in traditional peer groups.
POD ranking software is available for free from Free Software. Then you can purchase unbiased universes for $50 for U.S., or $75 for foreign and global. How can you not afford an accurate performance ranking?